FormalPara Key Summary Points

This guideline is written for all patients with chemotherapy-induced neutropenia and fever.

In this guideline a distinction is made between pediatric and adult patients, and adult patients were subdivided into standard- and high-risk neutropenic patients.

Recommendations for empirical antimicrobial therapy in this guideline are based on the prevalence of resistance in the Netherlands.

Further advice given includes the optimal duration of antimicrobial treatment, indications for treatment adjustment or streamlining, and the management of febrile neutropenic patients admitted to the intensive care unit.

Introduction

Summary and Rationale of Current Guideline

Fever is often the only sign of onset of infection in the neutropenic patient. In case of fever, prompt initiation of adequate empirical antimicrobial therapy reduces the risk of morbidity and mortality. To provide evidence-based recommendations for treatment of neutropenic patients with fever, we sourced all relevant clinical guidelines published since 2010 (Supplementary Material A). If there was no consensus in these guidelines, we performed a systematic search of the recent literature (2010–2020). This guideline aims to provide clinicians guidance in choosing the best antibiotic strategy for patients with chemotherapy-induced febrile neutropenia in the Netherlands. When available, recommendations in this guideline distinguish between high- and standard-risk episodes and between pediatric and adult patients. We want to stress that all these recommendations should be used in conjunction with clinical judgement. An individual clinical patient assessment is always important and, if deemed necessary, one can deviate from the guideline on the basis of clinical judgement.

Questions Answered in this Guideline

For this guideline a number of key questions were formulated. These questions were all separately investigated for patients with high-risk neutropenia (absolute neutrophil count (ANC) < 0.5 × 109/L neutrophils for > 7 days) and for those with standard-risk neutropenia (ANC < 0.5 × 109/L for ≤ 7 days). Questions were investigated separately for both children and adults.

  1. 1.

    For which patient groups is the current guideline written?

  2. 2.

    What are the most common microbiological causes of febrile neutropenia?

  3. 3.

    What is the most suitable empirical treatment for febrile neutropenia?

  4. 4.

    How is treatment adjusted in case of clinical or microbiological diagnosis?

  5. 5.

    What is the optimal duration of treatment for fever of unknown origin (FUO)?

  6. 6.

    What is the predictive value of surveillance cultures for infections with multiresistant bacteria?

  7. 7.

    What are the indications for removal of central venous catheters (CVC) in patients with febrile neutropenia?

  8. 8.

    What is the role for granulocyte colony-stimulating factor (G-CSF) in treatment of febrile neutropenia?

  9. 9.

    What additional investigations should be done to rule out an infection in patients with FUO?

Synopsis of Recommendations

Table 1 summarizes the recommendations.

Table 1 Synopsis of recommendations
Full size table

Purpose and Scope of this Guideline

The Dutch Working Party on Antibiotic Policy (Stichting Werkgroep Antibioticabeleid, SWAB), established by the Dutch Society for Infectious Diseases, the Dutch Society for Medical Microbiology, and the Dutch Association of Hospital Pharmacists, coordinates activities in the Netherlands aimed at optimization of antibiotic use, containment of the development of antimicrobial resistance, and limitation of the costs of antibiotic use. By means of the evidence-based development of guidelines, SWAB offers local antibiotic and formulary committees a guideline for the development of their own, local antibiotic policy. SWAB yearly reports on the use of antibiotics, on trends in antimicrobial resistance, and on antimicrobial stewardship activities in the Netherlands in NethMap (an annual report available from www.swab.nl), in collaboration with the Dutch National Institute for Public Health and the Environment.

Patients that suffer from neutropenia as a result of chemotherapeutic treatments are at high risk for infectious complications resulting in significant morbidity and mortality [1]. Fever may be the only clinical symptom at the onset of infection and should prompt rapid initiation of empirical treatment with broad-spectrum antimicrobial therapy. This treatment reduces the risk of death for patients with febrile neutropenia [2]. There are currently no Dutch national guidelines available to guide the choice of empirical antimicrobial therapy in this patient population, leading to a variety of empirical therapy approaches across the Netherlands [3].

This guideline aims to provide clinicians guidance in choosing the best antibiotic strategy for patients with febrile neutropenia.

Method

The guideline committee consisted of members delegated by their respective professional bodies, namely the Dutch Society for Infectious Diseases, Dutch Society for Medical Microbiology, Dutch Society for Hematology, Dutch Society for Medical Oncology, Dutch Association of Hospital Pharmacists, and Dutch Society for Pediatrics. No patient input was sought for the development of this guideline. After consultation with the members of these professional societies a number of questions and comments were raised and answered by the guideline committee and, when required, guideline text was adapted according to these comments. The revised guideline manuscript together with the point-by-point response have been approved by the SWAB and made publicly available on their website [4] (Fig. 1).

Fig. 1
figure 1

$For dosages in children, see www.kinderformularium.nl. *This dose differs from the EUCAST recommended therapeutic dose for treatment of invasive P. aeruginosa infection, for rationale see “Choice of Initial Empirical Antimicrobial Therapy/What is the Most Suitable Empirical Treatment for Febrile Neutropenia?” **3GCR: third-generation-cephalosporin resistance (e.g., due to production of AmpC or extended-spectrum beta-lactamases, ESBL). This is only relevant in case a cephalosporin is used. ***Skin: Gram-positive coverage (e.g., flucloxacillin); CVC: Gram-positive coverage including coagulase-negative staphylococci (CNS) (e.g., glycopeptide or oxazolidinones such as vancomycin or linezolid); neutropenic enterocolitis: anaerobic coverage (e.g., metronidazole). ****In case of neutropenic enterocolitis, no streamlining or discontinuation is advised except for addition of gram-positive coverage based on blood cultures. Reprinted with permission from Ned Tijdschr Hematol 2022;19(4):171–8

Flowchart for treatment.

Full size image

This guideline was developed according to the Dutch Antibiotic treatment Working Group (Stichting Werkgroep Antibioticabeleid, SWAB) tool guideline development and the AGREE-II tool for guideline development [5, 6]. The guideline committee consulted the European Committee on Antimicrobial Susceptibility Testing (EUCAST) breakpoints and their respective dosages for antimicrobial susceptibility. Empirical therapy advice was based on standard dosages that cover treatment of most pathogens, but often are not advised for therapy of invasive infections with Pseudomonas aeruginosa. In case clinical trials consistently used other dosages, these were advised (which was the case in imipenem–cilastatin, also see “Choice of Initial Empirical Antimicrobial Therapy/What is the Most Suitable Empirical Treatment for Febrile Neutropenia?”). Nine clinically relevant research questions with sub-questions were formulated on the basis of committee members’ clinical experience. The selection of questions is based on what the experts believe to be the most relevant questions encountered during the diagnosis and management of neutropenic patients with fever. The guideline committee deemed that answering these questions would aid physicians managing neutropenic patients with fever the most. Because this guideline focuses on the empirical antibiotic treatment of neutropenic patients with fever these questions include neither the diagnosis and treatment of invasive fungal infections nor questions concerning prophylaxis. In addition, for the management of invasive fungal infections we refer to the SWAB guideline on the management of invasive fungal infections [7].

As literature source, the committee used a selection of clinical guidelines that had been published since 2010, presented in Supplementary Material A. The recommendations concerning the preformulated research questions in these guidelines were compared to each other and provided the basis for this new SWAB guideline. Comparisons were made on three levels: the recommendation itself, the strength of the recommendation, and the level of evidence. Whenever source guidelines had high level of agreement, advice was adopted. Discrepancies between the guidelines lead to a new literature search.

For the review of the literature, references quoted in the respective guidelines were complemented with published articles on the subject found in PubMed up until January 1, 2020. Search terms were used (see Supplementary Material B for details) and all articles were screened on the basis of title and abstract for full text review. Full text review of selected articles was carried out by a subgroup of at least three people of the guideline committee, which led to a recommendation that was discussed in a plenary session by the full guideline committee and adopted after consensus was reached.

This guideline is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

For classification of the strength of the recommendation the GRADE (Grading of Recommendations Assessment, Development and Evaluation) system was used [8]. The GRADE system is a method of classifying quality of evidence and the strength of the accompanying recommendation. The strength of recommendations was graded as strong or weak, taking the quality of evidence, patients’ values, resources and costs, and the balance between benefits, harms, and burdens into account (Fig. 2). Quality of evidence is inherently linked to the strength of the recommendation: higher-quality evidence leads to more certainty on effect of the intervention (Table 2).

Fig. 2
figure 2

Approach to and implications of rating the quality of evidence and strength of recommendations using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology

Full size image
Table 2 GRADE [8]
Full size table

Results

Scope of the Guideline/for Which Patient Groups is this Guideline Written?

Chemotherapy-Induced Neutropenia

During neutropenic episodes, the innate immune response against microbial disease is largely attenuated and fever may be the sole symptom of a life-threatening infection. Although neutropenia may result from many different causes such as bone marrow failure, autoimmune disease, or congenital syndromes, the best recognized and studied causes of neutropenic episodes—during which fever should promptly be treated—result from myelosuppressive chemotherapy [1, 9, 10]. Treatment with these agents causes not only myelosuppression but may also result in mucositis. Febrile episodes in patients that suffer from the combination of a disrupted epithelial barrier in combination with lack of neutrophils have been extensively investigated. In contrast, no trials have been performed in febrile neutropenic patients in which neutropenia results from causes other than chemotherapy. Therefore, the recommendations given in this guideline are applicable foremost to the classical chemotherapy-induced neutropenia population. For neutropenic patients treated with agents that are not categorized as classical chemotherapeutical agents [e.g., but not limited to hypomethylating agents (HMA) or venetoclax] or in whom neutropenia results from hematological disease [e.g., but not limited to MDS, aplastic anemia, autoimmune or cyclical neutropenia or cytokine release as seen upon treatment with chimeric antigen receptor T cells (CAR-T cells)], no recommendations can be made on the basis of clinical trials, and treatment should be tailored individually.

To distinguish between high- and standard-risk neutropenic episodes, depth and duration of neutropenia are most often used. Often, high-risk patients receive prophylactic antibiotics, are hospitalized for the total duration of the neutropenic period for supportive treatment of cytopenias and mucositis, and are at higher risk for non-bacterial causes of infections such as invasive fungal disease. Whenever possible, advice in this guideline distinguishes between high- and standard-risk episodes. Moreover, when possible, recommendations distinguish between pediatric and adult patient populations.

Fever

In clinical guidelines and trials on the topic of febrile neutropenia, the definitions of fever and methods by which body temperature is measured are not consistent. Most consistently, fever is defined as a temperature measured orally of ≥ 38.3 °C measured once, or as ≥ 38.0 °C lasting for at least 1 h or measured twice within 12 h [11]. The guideline committee recognizes that a pragmatic approach of defining fever as a temperature of ≥ 38.5 °C at one time point is often employed and long-term experience with this approach has confirmed its safety. Because no unambiguous advice can be extracted from the available literature on the method by which temperature should be measured, we do not specify the method. However, it is common practice in many hospitals to use tympanic measurements.

High- and Standard-Risk Neutropenia

Pre-emptive risk stratification for infectious complications can be done by anticipating the depth and duration of neutropenia [12]. We utilized the following definition of high-risk versus standard-risk neutropenia in adults [11].

High-risk: absolute neutrophil count (ANC) < 0.5 × 109/L or an ANC that is expected to decrease to < 0.5 × 109/L over the next 48 h with an expected duration of neutropenia > 7 days.

Standard-risk: ANC < 0.5 × 109/L or an ANC that is expected to decrease to < 0.5 × 109/L over the next 48 h with an expected duration of neutropenia ≤ 7 days.

Patients assigned to the standard-risk group may exhibit individual characteristics, such as critical illness, justifying escalation of antibiotic treatment. We therefore propose different treatment for patients in which admission to the intensive care unit (ICU) is required for support of the febrile episode (see “Additional Treatment for Patients with Central Venous Catheters”).

In the absence of a generally accepted risk score for children and little data on oral outpatient treatment, there is no distinction between standard-risk and high-risk neutropenic episodes in children with FUO (Table 3).

Table 3 Synopsis of recommendations: question 1
Full size table

Most Common Microbiological Causes of Febrile Neutropenia

In case of fever in the neutropenic patient, microbiological documentation is only possible in 20–30% of the cases and blood cultures are positive in 10–25% with a bloodstream infection (BSI) incidence as high as 13–60% in myeloablative hematopoietic stem cell transplantation (HSCT) recipients [13,14,15]. In studies describing prevalence of bacteremia, patients were included with both fever of unknown origin as well as with fever in the context of clinically apparent foci [16,17,18,19,20,21,22,23,24,25,26]. These studies can thus be used to identify pathogens that are found in blood cultures of these patients, but specific prevalence and distribution in cases of fever of unknown origin (which is the most common cause of antibiotic treatment) are largely unknown.

Staphylococcus aureus is a rarely encountered pathogen during febrile neutropenia (0–3%, Tables 1 and 2, which includes patients with clinical symptoms other than fever) and infection is most often accompanied by clinical symptoms involving skin or central venous catheter. S. aureus is thus an infrequent cause of fever of unknown origin.

Most Common Microbiological Causes of Febrile Neutropenia in High-Risk Neutropenic Adult Patients

A summary of trials describing microbiological results of adult high-risk febrile neutropenic patients with and without antibiotic prophylaxis is presented in Table 1 [17,18,19,20,21, 27]. Gram-positive bacteria were most frequently (3–31%) identified in high-risk neutropenic patients, in all [17,18,19,20, 27, 28] but one study [21]. In comparison, Gram-negative bacteria were less frequently found. The proportion of patients with febrile neutropenia with Gram-negative pathogens in blood cultures differed between the group receiving antibiotic prophylaxis (with fluoroquinolones) compared to the group without prophylaxis; 1–8% in patients with and 4–13% in patients without antibiotic prophylaxis. Of the study patients, 0–4% had positive blood cultures for P. aeruginosa (Table 1).

Most Common Microbiological Causes of Febrile Neutropenia in High-Risk Neutropenic Pediatric Patients

In high-risk neutropenic pediatric patients, the same distribution of pathogens was found as in the adult patients described in the previous section. In a randomized controlled trial that included 617 children with high-risk neutropenia (198 children with acute leukemia and 419 children undergoing stem cell transplants) the likelihood of bacteremia between those receiving levofloxacin prophylaxis was compared to those without prophylaxis [16]. Gram-positive bacteremia was most frequent with viridans group streptococci as the most common pathogens. None of the children receiving levofloxacin prophylaxis developed S. aureus bacteremia. Prophylaxis with levofloxacin reduced the number of patients with Gram-negative bacteremia (GNB) from 34 without prophylaxis to 11 in the groups with prophylaxis (Table 4).

Table 4 Distribution of bloodstream isolates recovered from patients with or without ciprofloxacin or levofloxacin prophylaxis during neutropenia
Full size table

Most Common Microbiological Causes of Febrile Neutropenia in Standard-Risk Neutropenic Adult Patients

Likewise, a summary of microbiological data from trials describing standard-risk adult neutropenic patients with low risk for infectious complications, who were eligible for outpatient treatment, was made. In these studies, the definition of risk was not standardized. Most studies included patients with an estimated duration of neutropenia less than 7 days and low burden of disease (these patients had a high Multinational Association for Supportive Care in Cancer (MASCC) risk index (or MASCC score), or would be expected to have a high MASCC score) (Table 2) [22,23,24,25,26, 29].

In this standard-risk patient population with a low burden of disease (high MASCC score) P. aeruginosa (≤ 1.3%) and S. aureus (≤ 1.2%) BSIs are rare. Overall Gram-positive bacteria were more prevalent compared to Gram-negative bacteria in blood cultures from standard-risk patients, 1.6–6.4% versus 2.3–4.4% (Table 5).

Table 5 Distribution of bloodstream isolates recovered from standard-risk adult neutropenic patients
Full size table

Most Common Microbiological Causes of Febrile Neutropenia in Standard-Risk Neutropenic Pediatric Patients

In pediatric patients, no generally accepted definition exists to identify patients with a low risk for complications and pediatric studies included in the aforementioned meta-analysis all had different inclusion and exclusion criteria; some of those studies included only patients with negative blood cultures and are therefore were of little value. Therefore, microbiological data derived from studies in which high-risk neutropenic pediatric patients were included (as described in “Most Common Microbiological Causes of Febrile Neutropenia in High-Risk Neutropenic Pediatric Patients”) were used as background information for the recommendations for children.

Choice of Initial Empirical Antimicrobial Therapy/What is the Most Suitable Empirical Treatment for Febrile Neutropenia?

In patients without any sign of infection, prompt initiation of empirical therapy, awaiting blood culture results, is necessary to reduce mortality [2]. This therapy is focused on treating pathogens on the basis of prevalence and severity of disease caused. Pathogens that cause the highest risk of severe morbidity and mortality are Gram-negative bacteria. Although P. aeruginosa is rarely encountered in the current age of antibiotic prophylaxis, untreated this pathogen carries high morbidity and mortality. Moreover, all reference guidelines advise targeting this pathogen in empirical therapy. Thus, initial empirical therapy is foremost focused on adequate treatment of Gram-negative bacteria (including P. aeruginosa) with antipseudomonal beta-lactams. Empirical treatment advised in this guideline may differ from optimal therapeutic regimens for invasive infections with P. aeruginosa with respect to dose and mode of administration, but may be altered accordingly upon identification of this pathogen. Arguments for advised dose and mode of administration consist of toxicity, non-inferiority in randomized trials, and central venous lumen occupation. This is indicated in the following text.

The most encountered Gram-positive pathogens (coagulase-negative staphylococci (CNS), enterococci, and streptococci) most often do not cause a high burden of disease or overt sepsis, and additions to empirical therapy targeting these bacteria do not lead to better outcomes in non-septic patients [30]. Of these Gram-positive pathogens, viridans group streptococci may cause more burden of disease than CNS and enterococci and the need to empirically treat these bacteria is debated.

As already stated, occurrence of S. aureus in blood cultures is in most cases accompanied by additional clinical symptoms, for which additional considerations are described in “Should Empirical Antibiotic Therapy be Adjusted in Case of a Clinically Apparent Focus?” Thus, additional empirical antibiotic treatment will be initiated at time of treatment.

High-Risk Neutropenic Episodes

The practice of treating with antipseudomonal beta-lactam antibiotics dates from the 1960s when P. aeruginosa emerged as a common cause of BSI in the immunocompromised. Despite a declining incidence since, P. aeruginosa remains a serious cause of bacteremia with a very high mortality rate, ranging from 18% to 61% in neutropenic patients in more recent literature [31, 32]. When comparing antipseudomonal beta-lactam monotherapy treatments, the most recent Cochrane meta-analysis showed that—besides cefepime—carbapenems, ceftazidime, and piperacillin–tazobactam have comparative efficacy and toxicity and can all be used for febrile neutropenia [33]. Although all‐cause mortality was lower with piperacillin–tazobactam versus all other antibiotics, no statistical significant difference was found for infectious‐related mortality and clinical failure overall [33].

Cefepime

The possible excess mortality of cefepime demonstrated in an earlier meta-analysis was not confirmed by a data re-evaluation performed by the US Food and Drug Administration (FDA), which resulted in maintenance of the FDA approval for cefepime [34,35,36]. Difficulties with interpretation of the earlier mentioned meta-analysis included that although cefepime-treated patients had slightly but significantly increased mortality, no infection-related mortality difference was demonstrated. Moreover, the cefepime dose used in several of the studies was lower than the currently advised cefepime dose based on EUCAST. On the basis of this re-evaluation and extensive clinical experience, all but one of the reference guidelines have included cefepime as primary empirical treatment, with none recommending against. Cefepime is a fourth-generation cephalosporine with broad coverage of Gram-negative bacteria including P. aeruginosa and AmpC-carrying Enterobacterales such as Enterobacter spp. Moreover, cefepime is effective against streptococci (including streptococci with reduced penicillin sensitivity) and methicillin-sensitive S. aureus. It is not effective against anaerobic bacteria and ESBL-producing Enterobacterales. Even though cefepime has been used internationally for more than 25 years it has only recently been registered in the Netherlands for treatment of patients with fever and neutropenia and other indications. Several Dutch hospitals have adopted its use since.

Ceftazidime

Although initial empirical therapy is foremost focused on treating Gram-negative bacteria, the more limited coverage of Gram-positive bacteria by ceftazidime should be addressed, since no EUCAST breakpoints are provided for the treatment of S. aureus and streptococci. As stated previously, initial treatment of S. aureus is not required in patients without clinical symptoms indicating CVC or skin infection and initial treatment of streptococci is debated, since streptococcal infections, just as CNS or enterococcal infections, often have low clinical burden. Furthermore, the advised dosage of ceftazidime of 2000 mg q8h potentially provides adequate coverage of wild-type viridans streptococci based on pharmacokinetic/pharmacodynamic (PK/PD) data. In addition, treatment with ceftazidime was found to be non-inferior compared to piperacillin–tazobactam, cefepime, or carbapenems [33], and empirical addition of agents targeting Gram-positive bacteria (e.g., glycopeptides, beta-lactams, and others) did not result in better patient outcomes, although treatment failure (including requirement to start additional treatment upon identification of pathogens) was increased [30]. The combination of low virulence, antistreptococcal activity of ceftazidime, and clinical non-inferiority supports the recommendation of ceftazidime as a viable agent for the treatment of high-risk neutropenic patients.

Aminoglycosides

A large number of trials, summarized in a systematic meta-analysis, evaluated the use of aminoglycoside-containing combination therapy compared to antipseudomonal monotherapy. No advantage has been identified for the combination regimens, although toxicity emanating from these agents can occasionally be problematic [37,38,39]. For children with high-risk febrile neutropenia, intravenous monotherapy with antipseudomonal beta-lactams was found to be similarly appropriate [40].

Mode of Infusion

In non-neutropenic patients with sepsis, current guidelines advise extended or continuous infusion of specific beta-lactam antibiotics to optimize achievement of appropriate PK/PD targets [41]. It has been advocated that PK/PD targets may be higher in patients without alternative defense mechanisms, such as neutropenic patients [42], and administration by prolonged infusion may yield the highest chances of reaching the required targets. Moreover, in febrile neutropenic patients with hematological malignancies, certain underlying conditions may alter the PK of hydrophilic antibiotics such as beta-lactams, further compromising PD target attainment for P. aeruginosa and Enterobacterales using standard intermittent infusion regimens [43]. Administration by prolonged infusion may be imperative to reach the required PK/PD target, with both extended and continuous infusion having proven to be successful dosing strategies in PK studies with antipseudomonal beta-lactams [43,44,45,46,47]. Clinical data on effects of the beta-lactam infusion mode in neutropenic patients, however, are scarce. A retrospective study showed that 4-h extended infusion of meropenem led to better clinical outcome than conventional intermittent infusion [48]. It was independently associated with clinical success at day 5, fewer additional antibiotics, faster defervescence, and more rapid decrease of C-reactive protein but no differences in length of hospital stay or mortality were found. A randomized open-label trial performed in Israel has studied efficacy of extended infusion of ceftazidime and/or piperacillin–tazobactam versus bolus infusions in the neutropenic patient population. The extended infusion was superior in reaching a composite endpoint of clinical infectious response. No differences were found analyzing any of the single components of the outcome (defervescence, clinical failure, antibiotic switch, persistent BSI, mortality, length of hospitalization) [49]. Another study comparing extended (3 h) infusion of cefepime to standard 30-min infusion reported a shorter time to defervescence in neutropenic patients with fever receiving extended infusion, but no differences were found for clinical success, in-hospital mortality, length of hospital stay, and need for additional antimicrobials [50].

Currently, a multicenter, open-label, randomized, superiority clinical trial is being conducted in hematological neutropenic patients treated with cefepime, piperacillin–tazobactam, or meropenem to assess the clinical efficacy of extended versus intermittent beta-lactam infusion [51].

On the basis of the clinical evidence available, continuous or extended infusion treatment modalities are advised in septic patients. For non-septic patients, while awaiting further scientific evidence, mode of treatment infusion (bolus, continuous, or extended infusion) can preferably be advised. When using continuous infusion, a loading dose should be administered in order to rapidly achieve adequate serum concentrations.

Carbapenems

In an era of increasing antimicrobial resistance, restricting the use of carbapenems is considered good practice and antimicrobial resistance can be threatening on the population level as well as for the individual patient [52]. Benefits of carbapenems emanate from its broad antibiotic spectrum (including activity against third-generation cephalosporin-resistant (3GCR, e.g., AmpC and ESBL) Enterobacterales, methicillin-sensitive S. aureus, and viridans group streptococci, and the equal efficacy compared to other antipseudomonal beta-lactam antibiotics in the treatment of febrile neutropenia). The broad spectrum of carbapenems may result in reduced requirement of additional antibiotic agents that in turn could cause medication interactions or toxicity. Its disadvantages, encompassing collateral damage to the (intestinal) microbiome that is caused by the use of unnecessary broad-spectrum antibiotics, is increasingly recognized. In particular, use of carbapenems may be associated with selection of multidrug-resistant bacilli, predisposition to fungal infections, and development of Clostridioides difficile-associated diarrhea [53,54,55,56]. However, in addition to reduced prescription of carbapenem antibiotics, antibiotic stewardship depends on proper indication and timely discontinuation of antibiotics. Local bacterial epidemiology, prevalent resistance patterns, and patients’ risk factors for infection caused by resistant bacteria (e.g., ESBL colonization) should be taken into account when selecting an agent for empirical antibiotic therapy. On the basis of these considerations, a majority of the guideline committee members favored the recommendation of non-carbapenem agents (ceftazidime, cefepime, piperacillin–tazobactam) as a first choice for the treatment of neutropenic patients during high-risk episodes. Carbapenems (meropenem, imipenem–cilastatin) are the second choice. The advised dose of imipenem–cilastatine (500 mg/500 mg q6h) differs from that advised according to EUCAST for treatment of P. aeruginosa (1000 mg/1000 mg q6h). Reasons for this discrepancy are that the lower dose is most often used in clinical studies evaluating efficacy of imipenem–cilastatin, in which efficacy was equal to all other advised beta-lactams. In addition, increasing the imipenem–cilastatin dose may result in increased toxicity (most notably nephrotoxicity) while adequately targeting a larger proportion, but not all wild-type P. aeruginosa strains. These data caused the commission to advise a dose of 500 mg/500 mg q6h. Upon identification of P. aeruginosa in blood cultures, treatment should be altered accordingly.

In conclusion, we recommend to use any of the following beta-lactam antibiotic drugs with antipseudomonal activity for adult patients with FUO and high-risk neutropenia and all children with FUO: first choice—ceftazidime 2000 mg q8h; cefepime 2000 mg q8h; piperacillin–tazobactam 4000/500 mg q6h. Second choice—meropenem 1000 mg q8h; imipenem–cilastatin 500/500 mg q6h. Dosages for children should be altered according to age and weight (www.kinderformularium.nl). In adult patients who are extremely underweight or overweight, consultation of a pharmacologist for the appropriate beta-lactam dosage is advised.

Standard-Risk Neutropenic Episodes: Risk Assessment

For standard-risk neutropenic patients, oral and outpatient treatment can be considered if there is an individual low risk for serious complications. To aid risk identification for the individual patient the following risk scores are frequently recommended by international guidelines: MASCC risk index [57], the Talcott risk-scoring system [58], or the Clinical Index of Stable Febrile Neutropenia (CISNE). For patients with solid tumors, the CISNE is recommended, and some guidelines suggest performing CISNE scores in all patients in which MASCC scores indicate low risk for complications (ASCO/IDSA 2018) [59]. Although different risk scores may thus be used, most experience is obtained with the MASCC score, and a score of 21 or higher may support the notion that the patient is at a low risk of complications. Furthermore, trials using this score included patients with both solid tumors and hematological malignancies, making it a simple scoring method that can be performed in all emergency departments (the MASCC scoring system is available in a number of online calculators such as on mdcalc.com).

The ASCO guideline for pediatric patients with febrile neutropenia cited six different risk scores that rely on a single assessment at presentation and that have been validated in different pediatric populations, but were unable to clearly recommend any single prediction rule [60,61,62,63,64,65,66]. In addition, these scores were not used in trials examining oral outpatient treatment in children at low risk for complications. As a result of the absence of a generally accepted risk score for children and little data on oral outpatient treatment, all children with FUO should initially be treated with intravenous antibiotic agents.

Standard-Risk Neutropenic Patients with a Low-Risk of Serious Complications

For low-risk neutropenic patients (standard-risk neutropenia and a high (≥ 21) MASCC score), oral antibiotic treatment is safe. Several clinical trials have demonstrated equal efficacy of the combination of amoxicillin–clavulanate in combination with a fluoroquinolone in comparison to intravenous antibiotics [23, 67, 68]. In two trials, monotherapy with moxifloxacin was also shown to be safe and effective [24, 69] although moxifloxacin has no activity against P. aeruginosa [70, 71]. As a result of the exceedingly low prevalence of P. aeruginosa infections in this low-risk patient population (≤ 1%) and because patients with invasive P. aeruginosa infections will likely be identified by high burden of illness, there is no clear preference between moxifloxacin or the combination of amoxicillin–clavulanate plus ciprofloxacin [24]. In settings with a high prevalence of 3GCR Enterobacterales and fluoroquinolone resistance, inpatient treatment with a carbapenem should be considered in low-risk neutropenic patients [72]. In the Netherlands, national surveillance data (NethMap) on inpatient departments shows a background fluoroquinolone resistance of Enterobacterales and non-fermenters of 4–14% (ciprofloxacin resistance of E. coli, Klebsiella pneumoniae, Proteus mirabilis, Enterobacter cloacae complex, P. aeruginosa, and Acinetobacter spp.) and an estimated percentage of ESBL-carrying E. coli and K. pneumoniae of 6–9% [73]. Considering the Dutch antimicrobial resistance rates, both the combination of amoxicillin–clavulanate plus ciprofloxacin, or moxifloxacin monotherapy can be used in this population. In patients that have gastrointestinal complaints, a once-daily single pill regimen as with moxifloxacin may be regarded as more favorable, but drug interactions may cause prolonged QTc time.

Although fluoroquinolone prophylaxis is not advised in non-high-risk neutropenic patients that generally have short duration neutropenia, in selected cases, patients may still receive such treatments. Since all oral treatment regimens contain a fluoroquinolone, oral outpatient treatment is not recommended for patients in which fever develops during prophylactic treatment with fluoroquinolones. These patients should be regarded as at high risk for complications, and hospital admission and intravenous antibiotic treatment is advised.

In conclusion, we recommend to treat adult patients with FUO and standard-risk neutropenia and a high MASCC score, indicating low risk for serious complications, with the combination of amoxicillin–clavulanate 500/125 mg p.o. q8h plus ciprofloxacin 500 mg p.o. q12h, or with monotherapy moxifloxacin 400 mg p.o. q24h.

Standard-Risk Neutropenic Patients with a High-Risk of Serious Complications

Patients with standard risk neutropenic episodes that are at high risk for complications (e.g., MASCC score < 21) usually have a high burden of disease. Often, therapy that causes short-term neutropenia (< 7 days) results in mild mucositis and thus alternative foci that are at the root of their problems should be investigated. Epidemiology of pathogens in this patient group is elusive, since these patients are almost invariably excluded from trials and form a small subgroup. Often, these patients require medical support to such an extent that discharge is not possible and oral treatment in this group is not investigated nor advised. In patients that have standard-risk neutropenia, P. aeruginosa prevalence is low and a number of studies have evaluated safety and efficacy of alternative treatment regimens, such as with ceftriaxone monotherapy [74], or combination therapy with ceftriaxone and gentamicin [75], confirming safety in this patient population. No specific trials have been performed on the patient population with standard-risk duration (< 7 days), but with high risk for complications (MASCC < 21). Treatment with broad-spectrum antibiotics containing beta-lactams targeting Gram-negative bacteria, but as a result of the burden of disease also Gram-positive pathogens, is advised. This will be achieved by treatment with a regimen used for community-acquired sepsis. For adjustments based on clinically apparent foci in this population, see “Should Empirical Antibiotic Therapy be Adjusted in Case of a Clinically Apparent Focus?”

Additional Treatment for Patients with Central Venous Catheters

A number of trials summarized in a systematic meta-analysis [30] have shown that empirical addition of Gram-positive coverage using glycopeptides or addition of beta-lactam antibiotics directed against Gram-positive pathogens (e.g., flucloxacillin or amoxicillin–clavulanic acid) used for treatment of febrile neutropenia does not improve clinical outcome (defined as survival or infection-related mortality) at the cost of increased side effects. This only applies when there is no clear CVC entry infection. Patients included in the trials that were reviewed in this meta-analysis were not stratified according to the presence of a CVC, but the majority of the patients in the trials did have a CVC. Most bacteria associated with CVC infection that are insufficiently treated with single-agent beta-lactam regimens advised for febrile neutropenia are low-virulence organisms (CNS and enterococci) which do not require immediate empirical antimicrobial treatment. Treatment of these low-virulence pathogens can be initiated when identified from blood cultures. Therefore, additional Gram-positive coverage (e.g., but not limited to vancomycin) is reserved for settings in which infection of the CVC is clinically apparent. This recommendation does not apply to neutropenic patients admitted to the ICU, as these patients were not included in any of the trials included in the aforementioned systematic review [30].

Hemodynamically Unstable Neutropenic Patients/Neutropenic Patients Admitted to the ICU

Randomized controlled trials of neutropenic patients admitted to the ICU are lacking, and ICU referral is often a study endpoint. Therefore, recommendations are based on expert opinion. Moreover, most neutropenic patients that are hemodynamically unstable at presentation of fever have been excluded from clinical trials examining use of empirical antibiotic regimens. Although antipseudomonal beta-lactam monotherapy is the first choice for all high-risk neutropenic patients, guidelines commenting on the hemodynamically unstable (requiring relocation to the ICU) patients leave room for the addition of a second Gram-negative agent or a glycopeptide [13, 14, 52, 66, 76,77,78]. The IDSA guideline only recommends to broaden coverage for resistant Gram-negative bacteria in hemodynamically unstable patients with persistent fever after initial doses with standard agents for neutropenic fever [13].

Evaluation of the evidence for non-ICU patients showed that the addition of aminoglycoside, as described above, was not associated with better survival in high-risk neutropenic patients with fever. The routine addition of glycopeptides in high-risk neutropenic patients does not influence survival [30, 79]. Intravenous antipseudomonal beta-lactams remain the first-choice empirical therapy for children and high-risk neutropenic adult patients admitted to the ICU, and should be given without delay [78]. Although surveillance cultures adequately display colonization with resistant Enterobacterales and P. aeruginosa, these cultures may not have been routinely performed. Therefore, in order to target these bacteria (e.g., 3GCR Enterobacterales and P. aeruginosa) in patients with a lack of adequate surveillance cultures, potential escalation of the beta-lactam regimen, or addition of a second agent targeting Gram-negative bacteria may be considered on the basis of clinical grounds. Furthermore, in neutropenic hemodynamically unstable (requiring ICU admission) patients with a CVC, the addition of a glycopeptide or oxazolidinone (e.g., vancomycin, teicoplanin, linezolid) to treat possible CLABSI with CNS or enterococci may be considered, pending microbiological results. Empirical treatment for non-mold fungal infections (e.g., Candida spp.) can be considered in settings associated with increased prevalence of non-mold fungal infections: high-risk neutropenia without prophylaxis against fungal species or patients in which colonization with fungal species persist despite prophylaxis, especially when accompanied by mucositis. Starting treatment with empirical Candida-active agents (e.g., echinocandins) should only be considered in patients with high burden of disease (e.g., ICU admission, enterocolitis) in settings with high local incidence.

There is no evidence supporting a difference in the treatment of sepsis and septic shock in patients with neutropenia compared to non-neutropenic septic patients. We therefore recommend to treat adult patients with FUO and standard-risk neutropenia and a low MASCC score (indicating high risk for serious complications) as per the local treatment protocol for sepsis [41] (Table 6).

Table 6 Synopsis of recommendation: question 3
Full size table

How is Treatment Adjusted in Case of Clinical or Microbiological Diagnosis?

Should Empirical Antibiotic Therapy be Adjusted in Case of a Clinically Apparent Focus?

In the majority of febrile episodes in neutropenic patients, no specific origin can be identified. Nevertheless, fever should always prompt clinical evaluation including patient history and physical examination, since upon finding a potential infectious focus site, specific cultures may be taken and empirical antibiotic therapy may be altered. It should be taken into account that a clinically apparent infection in neutropenic patients may have other causative agents than in otherwise healthy patients (e.g., Gram-negative pathogens in skin infections [80]), and that omitting antibiotic treatment targeting Gram-negative bacteria may have an unfavorable outcome. Certain foci may require expansion of the spectrum of the initial empirical antibiotic regimen. For example, in skin infections, coverage of Gram-positive agents including S. aureus is warranted, especially in hospitals in which ceftazidime is the empirical treatment. For suspected urinary tract infections (UTIs) and pneumonia, no additional treatment is required, unless less common pathogens are suspected on clinical grounds (e.g., S. aureus pneumonia during influenza season, especially when ICU admission is necessary). Special care should be taken in case of a suspected central nervous system infection, and immediate consultation with a specialist should be initiated. Therapy should be targeted to treat a clinical apparent focus in clinically stable patients with resolution of fever after 48 h of initial empirical therapy as addressed as in “Choice of Initial Empirical Antimicrobial Therapy/What is the Most Suitable Empirical Treatment for Febrile Neutropenia?”, based upon the spectrum of microorganisms typically involved in the respective clinically documented infection.

Neutropenic Enterocolitis

Severe and prolonged neutropenia may result in reduced intramucosal defense against gut pathogens and enterocolitis may develop, often resulting in abdominal pain, diarrhea, and cecal wall thickening in combination with “fat stranding” on CT scan, a clinical syndrome known as neutropenic enterocolitis or typhlitis. Neutropenic enterocolitis is difficult to distinguish from or may be accompanied by enterocolitis caused by C. difficile, and the imminent diagnosis warrants testing for C. difficile in all patients [81, 82]. Anaerobes and Gram-negative organisms predominate as causative agents in neutropenic enterocolitis, and treatment regimens may consist of a combination of an antipseudomonal cephalosporin plus metronidazole, or monotherapy with piperacillin–tazobactam or a carbapenem [13]. Furthermore, vigilance for infections with yeast species is warranted for patients that suffer from neutropenic enterocolitis, see “Hemodynamically Unstable Neutropenic Patients/Neutropenic Patients Admitted to the ICU”.

Should Empirical Antibiotic Therapy be Streamlined or Adjusted Upon Retrieval of Possible Causative Pathogens from Blood Culture

Antibiotic streamlining encompasses altering the empirical broad-spectrum antibiotic treatment to specific and targeted treatment, in which narrowing of the antibiotic spectrum is pursued.

Although the quality of evidence is very low, guidelines are equivocal in advising that when a causative microorganism is identified, initial antimicrobial agents should be streamlined accordingly. When altering antibiotic therapy on the basis of positive blood cultures it is important to consider the etiologic relevance of the positive blood culture. Although Gram-negative bacteria are generally considered of etiologic relevance, the clinical relevance of Gram-positive bacteria is variable depending on the bacterial species identified and may result from contamination. Moreover, blood cultures may yield multiple findings (during high-risk neutropenia, polymicrobial findings range from 0 to 4.5% [27]). Therefore, caution is advised during early streamlining or altering antibiotic therapy in case of Gram-positive pathogens (Table 7).

Table 7 Synopsis of recommendation: question 4
Full size table

What is the Optimal Duration of Treatment for FUO?

In patients with FUO (defined as fever with a lack of microbiological or clinically documented infection), no definitive evidence on optimal duration of treatment has been published. Traditionally, prolonged treatment was proposed until resolution of neutropenia. This practice was based on the assumption that fever resulted from translocation of bacterial antigens through a damaged digestive tract. Once a focus for infection, repeated bacterial translocation would ensue [13, 83]. To date, the American and Korean guidelines adhere to this advice [13, 76] and propose that long-term experience with this strategy has resulted in confirmation of its safety and efficacy. More recently, antibiotic stewardship, bacterial resistance, and other negative implications of reducing microbiome diversity, such as possible long-term effects on graft versus host disease, have resulted in the tendency to shorten treatment courses. Several authoritative guidelines advocate this strategy [11, 52, 77, 84]. A number of studies, which were performed primarily in children, confirmed safety of stopping antibiotic treatment after defervescence after 48 h [85, 86]. Of note, only children that had low risk of infectious complications were included in these studies (no reasons for prolonged hospitalization, underlying cancer in remission) and these children mostly had diagnoses of which treatment would have resulted in low-risk neutropenia in adults, being reflected in the absence of mortality in these studies.

In adults with high-risk neutropenia, prophylactic antibiotic regimens will mostly be resumed upon discontinuation of empirical antibiotics, resulting in maintained antibiotic treatment for the duration of neutropenia in most patients with high-risk neutropenia. Several guidelines advise a treatment duration with empirical antibiotics of 5 days after defervescence [11, 13, 84], with little evidence-based support. A number of observational publications have advocated safety of a 3-day treatment course in patients that have become free of fever [87, 88] and a Spanish observational study showed that the vast majority of blood cultures become positive within the first 24 h, obviating the need for long-term treatment in order to cover pathogens that require long culture times [89]. A recently completed Dutch trial compared a 3-day treatment course with 9 days of treatment with meropenem. In this trial, antibiotics were also discontinued in patients that remained febrile. Results of this study have not been published. Presumed safety of short-term regimens in combination with a preference to treat as short as possible in order to reduce antimicrobial resistance led to the recommendation to discontinue empirical antibiotic treatment in stable patients if no fever persists. Although most guidelines advise discontinuation of empirical antibiotic treatment after 72 h in these patients, considering the fact that a very small proportion of blood cultures will yield additional findings after 24 h of culture, stopping empirical treatment after 48 h is advised.

In patients that remain febrile, discontinuation of empirical antibiotic treatment is under increased scrutiny. Outside the aforementioned unpublished Dutch trial, no data underlie treatment advice. In patients in which antibacterial prophylaxis is given, reverting to this prophylactic regimen may be prudent in clinically stable patients that remain hospitalized with the goal of reducing treatment duration of broad-spectrum empirical antibiotics and complications resulting from these agents (e.g., C. difficile infections, candidemia) [90, 91]. Patients that are not treated with broad-spectrum empirical therapy and remain febrile should remain under close scrutiny, since other symptoms than fever (e.g., frank rigors or hypotension) should prompt re-initiation of empirical antibiotic treatment. Patients with persistent fever that is not responsive to empirical antibiotic treatment have a worse prognosis than patients in which fever abates, and in these patients, other infectious causes should be considered (e.g., but not limited to hepatosplenic yeast infections, invasive mold infections) (Table 8).

Table 8 Synopsis of recommendation: question 5
Full size table

What is the Predictive Value of Surveillance Cultures for Infections with Resistant Bacteria?

In previous studies, the sensitivity of colonization with multidrug-resistant (MDR) bacteria for MDR-BSI in the hematologic patient population ranged from 45% to 91% [92,93,94,95,96,97,98,99], with most evidence for and a very high negative predictive value of ESBL-E colonization for ESBL-E bacteremia (73.9–99.8%) [92, 94, 95, 97, 98]. Two studies showed that P. aeruginosa colonization independent of resistance can be predictive for infection [96, 100]. The ECIL-4 guidelines conclude that colonization or infection by resistant organisms is the most important risk factor for infection with resistant pathogens [52]. Adjustment of treatment based on colonization with specific pathogens or the selection of narrow-spectrum empirical antibiotic therapy based on the absence of (resistant) pathogens in routine surveillance cultures has not been studied. Most Gram-negative bacteria are covered by the empirical antibiotic therapy recommended by this guideline (“Choice of Initial Empirical Antimicrobial Therapy/What is the Most Suitable Empirical Treatment for Febrile Neutropenia?”). When patients are colonized with 3GCR Enterobacterales or P. aeruginosa, i.e., resistant to the used empirical agents, empirical antimicrobial treatment should be adjusted accordingly. Carbapenem-resistant Enterobacterales or P. aeruginosa are still very rare in the Netherlands but studies from countries with high background resistance rates (e.g., Italy and India), demonstrate the association between colonization and infection with these very resistant bacteria [101,102,103,104,105]. These studies also demonstrated a significant association between carbapenem-resistant Gram-negative bacteria (4/5 studies included only Enterobacterales) and mortality. We therefore recommend to adapt empirical treatment in patients colonized with these bacteria. As a result of limited data and possible lower virulence and weak directly attributable mortality, non-fermentative Gram-negative bacteria (other than P. aeruginosa) resistant to the empirical treatment regimen (e.g., Acinetobacter spp.) are not included in these recommendations and should be discussed per individual case [106].

Initial empirical treatment does not include the coverage of vancomycin-resistant enterococci (VRE), penicillin-resistant viridans streptococci, and/or Candida species. VRE colonization is found to be predictive of VRE infection in several studies [95, 107,108,109,110,111], but enterococci are not covered in empirical antibiotic regimens for febrile neutropenia because they are of low pathogenicity. Therefore, the adjustment of antibiotic therapy because of VRE colonization is only recommended when infection with enterococci is highly suspected, or in critically ill patients (e.g., ICU admission, see “Hemodynamically Unstable Neutropenic Patients/Neutropenic Patients Admitted to the ICU”). Evidence for the relationship between colonization and infection with penicillin-resistant viridans streptococci is scarce and no evidence-based recommendations can be made [112, 113]. Colonization with Candida species, especially multisite colonization, is a risk factor for candidemia or invasive candidiasis [114,115,116]. However, incidence of candidemia and/or invasive candidiasis is low and therefore the coverage of Candida species is not included in the empirical antimicrobial therapy recommended by this guideline (“Most Common Microbiological Causes of Febrile Neutropenia”). Initiating empirical antifungal therapy may result in excess costs and treatment‐related toxicities that may not be justified. Therefore, empirical therapy with antifungal agents is not recommend. Pre-emptive antifungal therapy should be considered in patients with high burden of disease (e.g., ICU admission, enterocolitis) in combination with one or more of the following (“Standard-Risk Neutropenic Episodes: Risk Assessment”) (Table 9):

  • Persistence of yeast species in surveillance culture

  • Absence of antifungal prophylaxis

Table 9 Synopsis of recommendation: question 6
Full size table

What are the Indications for Removal of CVC in Patients with Febrile Neutropenia?

All foreign bodies carry the risk of being a source for colonization and infection and consequently may cause fever. CVCs should be evaluated for potential site of infection in a febrile episode. In all patients, CVC removal is advised if there is no medical requirement.

Five trials specifically involving neutropenic patients with CVCs have been published [117,118,119,120,121]. In none of these studies was CVC removal versus maintenance investigated in the setting of a putative CVC infection. Therefore, the recommendation on CVC maintenance versus removal and CVC salvage using antimicrobial treatment is adopted from the IDSA guideline on catheter-related infections in immunocompetent patients [122, 123]. Risk balance between recurrence of BSI and removal of CVC should be made in all patients with a CVC. A lower threshold of CVC removal in neutropenic patients that have had a Gram-negative bacteremia or who are critically ill is justified. Immediate CVC removal is indicated for bacteremia with P. aeruginosa, S. aureus, and Candida species as per the central line-associated BSI (CLABSI), S. aureus bacteremia [124], and candidemia guideline [125] (Table 10).

Table 10 Synopsis of recommendation: question 7
Full size table

What is the Role for G-CSF in Treatment of Febrile Neutropenia?

In neutropenic patients that suffer from fever, reducing the duration of neutropenia may reduce the duration of the febrile period and aid in the treatment of febrile patients. To this end, treatment with granulocyte colony-stimulating factor (G-CSF) has been evaluated in patients with cancer in a number of randomized controlled trials, largely summarized in a systematic review [126]. In these studies, febrile patients were treated with antibiotics and with G-CSF, in contrast with treatment with antibiotics alone. These studies equivocally exhibited reduced length of neutropenia without beneficial effects on mortality. Although these studies were powered to evaluate use in specific infections (e.g., mold infections), the guideline committee advises against standard use of G-CSF as adjunctive treatment in febrile neutropenia (Table 11).

Table 11 Synopsis of recommendation: question 8
Full size table

What Additional Investigations Should be Done to Rule Out an Infective Focus in Patients with Febrile Neutropenia of Unknown Origin?

The initial diagnostic approach of the neutropenic patient with fever aims to establish a clinical and microbiologic diagnosis, which leads to targeted (antibiotic) treatment and thereby improving the patient’s prognosis. In neutropenic patients with fever, this should at least include clinical history, physical examination, and the drawing of blood cultures before antibiotic therapy is administered (peripheral and CVC).

Imaging

In patients with clinical signs and symptoms of pneumonia radiographic imaging (conventional chest X-ray radiography (CXR) or chest CT scan) is recommended and should be obtained within 24 h. In one study, sensitivity, specificity, positive predictive value, and negative predictive value for conventional radiography were 36%, 93%, 50% and 88%, and for low-dose CT scan 73%, 91%, 62%, and 94%, respectively [127]. Therefore, chest CT scan is the preferred modality owing to the higher sensitivity and specificity [127, 128]. The optimal timing of radiological imaging is not known; in studies and in clinical practice, CXR or chest CT scan is often performed within 24 h [127, 129].

In asymptomatic children, previous studies show that chest radiography rarely shows a pneumonia, and if CXR was not obtained no significant adverse clinical consequences were observed [40, 130,131,132]. The lack of consequence of the rare abnormal CXR in the absence of respiratory symptoms/signs has been confirmed in adults [129, 133]. Therefore, routine radiography in the workup of febrile neutropenia (CXR or chest CT scan) without symptoms of a respiratory infection is not recommended. This advice specifically concerns radiography in the first 24 h of fever and does not involve chest imaging aimed at diagnosing invasive fungal infections in patients with persistent fever.

Urine Analysis

During neutropenia, the diagnosis of a UTI can be challenging, as pyuria is not a reliable parameter in neutropenic patients with UTI [134]. In addition, UTI symptoms can be atypical or even absent [135], while a positive culture may reflect contamination of colonization instead of infection. However, for the diagnosis of a UTI, a positive urine culture combined with the clinical suspicion of a UTI remains the gold standard. Furthermore, routine urine analysis in the absence of complaints may result in excessive invasive procedures (as catheterization may be required in children) or therapeutic delay in the absence of therapeutic consequences.

In conclusion, routine urinalysis or urine cultures are not beneficial in patients that do not exhibit urinary tract complaints and may be unnecessarily invasive (e.g., requiring catheterization). Therefore, in both children and adults, urine cultures are recommended only when UTI is suspected or if the patient has a history of recurrent UTIs (Table 12).

Table 12 Synopsis of recommendation: question 9
Full size table

Conclusion

In this guideline we provided evidence-based recommendations using all clinical relevant guidelines published since 2010 as a source, supplemented with systematic searches and evaluation of the recent literature (2010–2020, Supplementary Material) and, where necessary, supplemented by expert-based advise.

Knowledge Gaps

The current guideline was reviewed by members of the professional bodies of the Dutch Society for Infectious Diseases, Dutch Society for Medical Microbiology, Dutch Society for Hematology, Dutch Society for Medical Oncology, Dutch Association of Hospital Pharmacists, and Dutch Society for Pediatrics. From these reviews and from the peer-review process of Infectious Diseases and Therapy a number of questions and limitations were raised of which some can be established as knowledge gaps. Some of these knowledge gaps may be used to formulate novel research questions when the guideline will be revised. The following knowledge gaps deserve attention in research in the coming years:

  • What are the prevalence, distribution, and resistance of pathogens that cause infection in patients with a standard-risk neutropenia with a high-risk of infectious complications (low MASCC score)?

  • What are oral alternatives for patients with a standard-risk neutropenia and a low-risk of infectious complications (e.g., levofloxacin or ciprofloxacin monotherapy or ciprofloxacin–amoxicillin combination therapy)?

  • What is the association between colonization with third-generation cepaholosporin-resistant Gram-negative bacilli and infection? How long after the identification of colonization is the risk of infection with these bacteria increased?

  • What are the causative patients with fever due to a clinical focus (e.g., cellulitis and pneumonia)?

  • Does reserving carbapenems in high-risk neutropenic patients with fever result in a reduction of antibiotic use?

  • Does discontinuation of antibiotic therapy in patients with persistent fever without a focus of infection lead to a reduction of antibiotic use?

  • What are the reasons to restart antibiotics after initial discontinuation of empirical antibiotic therapy after 48 h; in febrile neutropenic patients without a focus of infection and with negative blood cultures?

  • What actions should be taken (diagnostic or therapeutic) in patients without apparent clinical or microbiological focus with persistent fever despite empirical antibiotic treatment?

  • What are the indications for CVC change in patients with fever and a medical requirement of the CVC?

  • Are there biomarkers that may guide empirical management of febrile neutropenia?